Marvin 5.3.4 performance

I saw in your "history of changes" that you released Marvin 5.3.4 yesterday.  I emptied my Java cache and went to your demo page to try it out.  It took over 20 seconds for Marvin to launch.  When I pressed on the Tools menu, it took 5 seconds for a menu to appear.  When I chose About MarvinSketch, it took 7 seconds for "Module loading" to appear, then Marvin went blank, and it took an additional 23 seconds for both the About MarvinSketch text to come up and for the Marvin window to reappear.  I am using Safari 5.0.  

Even worse: I quit Safari, then went back to the Marvin demo page.  I clicked on the cyclopentane ring, then clicked on the Marvin drawing pad.  After no reaction for about 20 seconds, the beach ball appeared, and I had to force-quit Safari.

I think you will understand why when I say I am not going to upgrade to Marvin 5.3.4.  I'm disappointed that such serious problems are still present even in the fourth bugfix release.  

Hi Bob,

I've just downloaded Safari 5, and installed on my computer (Win 7), then checked the speed of loading the MSketch tryout page. It loaded in 5 seconds, and all of the menu items or toolbar icons responded quickly, even when trying it first time.

I guess at the moment when you tried it, the commmuication speed to the sever where the demo page is located was slow. Would you please try again?

Thanks and best regards,

Akos

I am now on the same computer (Safari 5.0, MacOS 10.5.8) at the University.  It took about 17 seconds for Marvin to load.  This time it did not crash when I drew cyclopentane, but it still took quite a while (10-15 seconds) to react.  When I went to Topics, the menu loaded immediately, but then the Module is loading window appeared, and nothing happened.  It wasn't until I closed the Module is loading window over a minute later that the Help message appeared.

Thank you, we'll try to figure out on a Mac computer what can be the problem.

Best regards,

Akos

I loaded Marvin 5.3.4 onto our local server here.  The performance problems are again very bad with Safari 5.0 and FIrefox, though Firefox is not as bad as Safari, and neither performs as badly with the local installation as they do with yours.

OK, we will cmpare performances.

Hi,

the issue might not only be Safari related, but a network issue. The reason for the first install the applet has to talk to

enternet.hu and also talks to Verisign. So depending on your ping statistics and connection to enternet.hu from

uky.edu you might experience some strong delays.

 

Anyway I would not touch Safari, I use Chrome and it works like charm, no

crashes, quite performant, but I had the same "cyclopropane" delay, but just for the few seconds due to the reasons above.

Ones its cached no problemo here in California. Check your routing with http://myspeed.visualware.com/

There is a big delay after it jumped over the ocean and leaves Franfurt, Munich. So actually a load balancer that

redirects all american traffic to the americas would be the correct solution. Anyway I am quite happy with the

internet speed compared to two-three years ago. And I had no crashes with MarvinSketch 5.3.4.

--------------------------------------------------------------------------------------------
MyConnection Server Connection Test to nas-96-1-e1.enternet.hu
        Performed on 25 Jun 2010 
--------------------------------------------------------------------------------------------

------------------------------------------------------------------------------------------------------------------------------------------------
| Hop | %loss | IP Address      | Node Name                                | Location               | ms  | Graph      | Network               |
------------------------------------------------------------------------------------------------------------------------------------------------
| 0   | 0     | 216.7.123.1     | vl104.core1.co.dorris.ca.cot.net         | -                      | 5   | +-         | Cal-Ore Telephone Co. |
| 1   | 0     | 216.7.123.1     | vl104.core1.co.dorris.ca.cot.net         | -                      | 0   | +          | Cal-Ore Telephone Co. |
| 2   | 0     | 216.7.112.187   | po1.vl298.edge.co.dorris.cot.net         | -                      | 0   | +          | Cal-Ore Telephone Co. |
| 3   | 0     | 216.7.112.186   | po1.vl298.edge.oithut.kfls.or.cot.net    | -                      | 0   | +          | Cal-Ore Telephone Co. |
| 4   | 0     | 207.109.242.237 | eug-edge-04.inet.qwest.net               | -                      | 4   | +          | -                     |
| 5   | 0     | 205.171.150.37  | eug-core-02.inet.qwest.net               | -                      | 4   | +          | -                     |
| 6   | 0     | 67.14.34.38     | sjp-brdr-04.inet.qwest.net               | -                      | 20  | +-         | -                     |
| 7   | 0     | 154.54.12.189   | te3-5.mpd01.sjc03.atlas.cogentco.com     | San Jose, CA, USA      | 19  | +          | -                     |
| 8   | 0     | 154.54.6.81     | te3-2.mpd01.sjc01.atlas.cogentco.com     | San Jose, CA, USA      | 23  | -+-        | -                     |
| 9   | 0     | 66.28.4.149     | te2-4.mpd01.sfo01.atlas.cogentco.com     | San Francisco, CA, USA | 20  | +          | -                     |
| 10  | 0     | 154.54.6.37     | te0-0-0-2.mpd21.mci01.atlas.cogentco.com | Kansas City, MO, USA   | 65  |    +       | -                     |
| 11  | 0     | 154.54.2.234    | te0-0-0-3.mpd21.ord01.atlas.cogentco.com | Chicago, IL, USA       | 68  |    +       | -                     |
| 12  | 0     | 154.54.30.233   | te0-0-0-7.ccr21.ymq02.atlas.cogentco.com | -                      | 90  |     +      | -                     |
| 13  | 0     | 154.54.28.94    | te0-0-0-4.mpd21.fra03.atlas.cogentco.com | Frankfurt, Germany     | 187 |         +  | -                     |
| 14  | 0     | 130.117.48.70   | te3-1.ccr01.muc01.atlas.cogentco.com     | Munich, Germany        | 193 |         +- | -                     |
| ?   | 100   | ...             | -                                        | -                      | -   |            | -                     |
| ?   | 100   | 62.112.195.61   | nas-87-1-l1.enternet.hu                  | -                      | -   |            | -                     |
------------------------------------------------------------------------------------------------------------------------------------------------

And the Kentucky Regional Optical Network (KyRON) unites UK and the
University of Louisville to the main Internet connector in Louisville
that links Kentucky to the rest of the world.

http://www.research.uky.edu/odyssey/summer09/internet2.html (so that should be fine).

 

Tobias

Marvin 5.2 continues to work perfectly well, so a change in Marvin 5.3 must have caused the performance problems.  Whatever that change is, it needs to be removed.  If the problem is really due to Marvin reaching across the network, then either it wasn't necessary for Marvin 5.2 to do that, so it's not necessary for Marvin 5.3, or it worked much better in Marvin 5.2.  Either way, the ChemAxon programmers need to fix the problem.  

BTW, Marvin is not validated on Chrome.  See this page.

Hi,

well I know lots of people work with Mac, so I guess its kinda important, but why not use

the real deal: start Fusion and work with WIN7 ... just kidding :-) hope you are not an Apple FanBoy.

Anyway here are my detailed observations from WIN XP 32 with JAVA 1.6 and MarvinSketch 5.3.4

I had no problems in terms of speed and all tools worked normal,

however the applications where throwing some errors here and

there. I used the Marvin Demo Page http://www.chemaxon.com/marvin/sketch/index.jsp

62.112.195.93
route to nas-96-1-e1.enternet.hu [62.112.195.93]
Ping statistics for 62.112.195.93:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 219ms, Maximum = 283ms, Average = 243ms

 

FireFox 3.6, WIN 32, applet and tools work ok

FireFox 3.6 (WIN32)
Java Plug-in 1.6.0_20
Using JRE version 1.6.0_20-b02 Java HotSpot(TM) Client VM
 

----
The publisher cannot be verified by a trusted source. Code will be treated as unsigned.
Name: OpenURL
java.lang.RuntimeException: java.lang.InteruptedException
----
Invokes
c:\xxx\system32\cscript.exe
----
Firefox 3.6 can not incorporate with current JAVA-plugin.
Probably its an old JAVA version update to a newer one.



ava.lang.RuntimeException: java.lang.InterruptedException
    at com.sun.deploy.security.TrustDecider.isAllPermissionGranted(Unknown Source)
    at sun.plugin2.applet.Plugin2ClassLoader.isTrustedByTrustDecider(Unknown Source)
    at sun.plugin2.applet.Plugin2ClassLoader.getTrustedCodeSources(Unknown Source)
    at com.sun.deploy.security.CPCallbackHandler$ParentCallback.strategy(Unknown Source)
    at com.sun.deploy.security.CPCallbackHandler$ParentCallback.openClassPathElement(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath$JarLoader.getJarFile(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath$JarLoader.access$700(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath$JarLoader$1.run(Unknown Source)
    at java.security.AccessController.doPrivileged(Native Method)
    at com.sun.deploy.security.DeployURLClassPath$JarLoader.ensureOpen(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath$JarLoader.<init>(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath$3.run(Unknown Source)
    at java.security.AccessController.doPrivileged(Native Method)
    at com.sun.deploy.security.DeployURLClassPath.getLoader(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath.getLoader(Unknown Source)
    at com.sun.deploy.security.DeployURLClassPath.getResource(Unknown Source)
    at sun.plugin2.applet.Plugin2ClassLoader$2.run(Unknown Source)
    at java.security.AccessController.doPrivileged(Native Method)
    at sun.plugin2.applet.Plugin2ClassLoader.findClassHelper(Unknown Source)
    at sun.plugin2.applet.Applet2ClassLoader.findClass(Unknown Source)
    at java.lang.ClassLoader.loadClass(Unknown Source)
    at java.lang.ClassLoader.loadClass(Unknown Source)
    at java.lang.ClassLoader.loadClass(Unknown Source)
    at sun.plugin2.applet.Plugin2ClassLoader.loadCode(Unknown Source)
    at sun.plugin2.applet.Plugin2Manager.createApplet(Unknown Source)
    at sun.plugin2.applet.Plugin2Manager$AppletExecutionRunnable.run(Unknown Source)
    at java.lang.Thread.run(Unknown Source)
Caused by: java.lang.InterruptedException
    at java.lang.Object.wait(Native Method)
    at java.lang.Object.wait(Object.java:485)
    at com.sun.deploy.util.DeployLock.lock(Unknown Source)
    ... 27 more

-----
tools stereoisomer and any other tool invokes:


Application path: c:\XXXX\ChemAxon\InstantJChem\instantjchem\modules\ext\marvinOLEServer.exe
Description: MarvinOLE server
File version: 2.2
Product name: MarvinOLE server

....

Application path: c:\XXXX\ChemAxon\InstantJChem\instantjchem\modules\ext\marvinOLEServer.exe
Description: MarvinOLE server
File version: 2.2
Product name: MarvinOLE server

 

Internet Explorer 8, WIN32, applet and tools work ok

Internet Explorer 8.06 (WIN32)
No problems, 
any tool invokes:


Application path: c:\XXXXX\ChemAxon\InstantJChem\instantjchem\modules\ext\marvinOLEServer.exe
Description: MarvinOLE server
File version: 2.2
Product name: MarvinOLE server


--------
closing IE gives errro:
Webpage error details

User Agent: Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; .NET XXX)
Timestamp: Fri, 25 Jun 2010  


Message: Object doesn't support this property or method
Line: 15
Char: 1
Code: 0
URI: http://www.chemaxon.com/marvin/sketch/index.jsp />

 

Google Chrome 4.1, WIN32, applet and tools work ok

Chrome 4.1.249 (WIN32)
invokes

Application path: c:\XXXXXXX\system32\cscript.exe
Description: Microsoft (R) Console Based Script Host
File version: 5.7.0.18066
Product name: Microsoft (R) Windows Script Host

Any tool invokes
Application path: c:\XX\ChemAxon\InstantJChem\instantjchem\modules\ext\marvinOLEServer.exe
Description: MarvinOLE server
File version: 2.2
Product name: MarvinOLE server

Cheers

Tobias

Just a notes for the attached reports.

In the log message, there is a warning OpenURL was treated as unsigned.

marvin.js call this code to detect before startup of
Marvin applet that the Java Plugin is compatible with the current
browser.

Since 5.3.5, OpenURL is going to be eliminiated. We have built in a most effective mechanism to check plugin compatibility.

I have the same performance problems in Marvin 5.3.5 that I have had in earlier iterations of Marvin 5.3 but that are completely absent in Marvin 5.2.6.

5.3.5 is a hotfix release. It does not contain any changes  that would influence the performancy of the applet.

Should I expect these problems to be fixed in Marvin 5.3.6?

Significant changes can not be expected. But we have eliminated a couple of insufficent requests between the applet and the server.